Skip to Content

Planning for a Climate-Changed World

As the global picture grows grimmer, states and cities are searching for the fine-scale predictions they need to prepare for emergencies–and to keep the faucets running.

On December 11, 1992, a powerful northeaster coalesced off the eastern seaboard of the United States, and an eight-foot storm surge struck New York City. Seawater swamped the Brooklyn Battery Tunnel to a depth of six feet, cascaded down PATH subway stairs in Hoboken, NJ, and forced LaGuardia Airport and many roads and subways lines to close. Had the storm been slightly stronger, a 10-foot surge could have devastated a far wider region, inundating low-lying areas like Coney Island and Manhattan’s financial district and overwhelming the 14 sewage plants dotting the New York City coastline.

Hot in New York, hotter in Jersey: Global warming is an abstraction; people want to know how warm their own cities or counties will get. That’s key to everything from agriculture to emergency planning. Global models typically provide estimates at a resolution of 150 kilometers. A regional model run inside a global one, however, can account for local topographical effects. The image above was produced by one such model: it shows average summertime surface temperatures in the 2050s, across counties in the New York metropolitan area. Knowing that it might be hotter in Manhattan than in the outer boroughs, and hotter still in parts of New Jersey, is vital for estimating, for example, the scale of health crises brought on by heat waves and ozone.

A flood of comparable height in New York City’s environs should occur about once every 100 years, on average, in the estimation of one Columbia University study. But global warming and rising sea levels–as well as the possibility of more-intense precipitation, stronger storms, and altered storm trajectories–will make such disasters more frequent. And to protect the people who live and work where disaster threatens, the critical first step is to determine how quickly and by how much, exactly, the threat is increasing. That knowledge is essential to deciding how seriously to consider specific countermeasures; for New York, these could range from mandatory evacuation plans for seaside neighborhoods to multibillion-dollar storm-surge barriers spanning the Verrazano Narrows and other key channels.

But there are no clear answers, and part of the problem is that well-documented predictions about planetary change haven’t generally been broken down in local terms. Though the Intergovernmental Panel on Climate Change (IPCC) has concluded with 90 percent certainty that human activity is warming the planet–and spelled out the likelihood of consequences that include higher seas, droughts, and fiercer storms–the United States is committing scant resources to providing usable information to the people who respond to emergencies, plan for urban development, manage coastal areas, and make sure the crops keep growing and the reservoirs stay full. “The challenge is to increase our capability to accurately forecast climate at the regional level,” says ­Ronald Prinn, an atmospheric scientist who directs the Center for Global Change Science at MIT. “That is what is needed in order to improve the information that government agencies get–[and] to then translate those regional forecasts into something useful at the city [or] state level.” Equipping people to deal with climate change could mean simply giving state and local planners access to a wealth of existing information–such as calculations made by the National Oceanic and Atmospheric Administration (NOAA) that could indicate how far inland storm surges would move if sea levels were higher. But it will also mean sharpening local and regional models, so that they can predict the effects of climate change in far greater geographic detail. And it will require new approaches to emergency planning, water-­supply management, and more.

Global warming will affect different regions in different ways. In 2000, a national assessment by the U.S. Global Change Research Program (USGCRP) warned generally about potential climatic changes in what it called “mega-regions” of the nation. “What we were able to do at that point was very limited,” recalls Michael MacCracken, an atmospheric physicist, now retired, who coördinated the assessment effort. And the study’s climate scenarios were based only on global models: “We really wanted to have more models, and more regional results, but we had very little resources to get that done.” Similar, very general statements about climate change across large regions appeared in the most recent IPCC assessment, the first time the IPCC has narrowed its focus even that much. The report pointed out, for example, that the southwestern United States will probably get even more parched than it is now. But what we need are projections on a far finer scale. With federal climate-science budgets cut to the bone in recent years, a few state and local governments are funding their own efforts in New York, California, and western states eyeing dwindling water supplies with alarm.

A Wet New York City

Rushing into her office near Columbia University, Cynthia Rosenzweig was chipper despite her evident exhaustion. An agronomist by training, she directs the Climate Impacts Group at NASA’s Goddard Institute for Space Studies (GISS) and advises New York City’s government on how climate change will intensify heat waves, stress upstate watersheds, and increase the risk of a devastating storm surge. She had just returned from Delhi, India, where she cowrote a summary of the 2007 IPCC reports, the first of which was released in February. Brightly painted papier-mâché elephants she’d brought back from her trip were arranged on the coffee table in her sixth-floor office overlooking 112th Street (as it happens, some of the highest ground in Manhattan). She sat down and, on her computer screen, called up images from a GISS global climate model.

Honed by a broad range of climate scientists, the model represents atmospheric and oceanic systems. Like other global models, it simulates interrelated processes: for example, the warming of Earth’s surface by solar radiation; the absorption of heat by the oceans; the reflection of solar energy by land surfaces, ice sheets, and particulates in the atmosphere; and the effects of the accumulation of excess carbon dioxide and other atmospheric gases that trap heat. Researchers test the accuracy of such models by seeding them with, for example, data on actual greenhouse­-gas emissions over the past 30 years and then seeing whether they return results consistent with temperature and other measurements recorded over that period. The goal, of course, is a model that can predict how much temperatures will continue to rise given various future greenhouse-gas emission levels, and how other parts of the climate system are likely to respond.

But the limitations of global models quickly become clear when Rosenzweig zooms in on a map of the eastern United States showing climate predictions for the 2050s. On the screen, a line cuts from eastern Pennsylvania to western Massachusetts. The area north of the line is yellow, representing a 2 ºC increase over historical averages; the area south of the line is more orange, indicating a 2.25 ºC increase. The entire New York metropolitan area, Connecticut, and much of Massachusetts and New Jersey are lumped together under a single temperature estimate. The same goes for several other variables, such as precipitation and evaporation rate. The problem is that one “grid box” in the typical global climate model–think of it as a pixel in a photograph–is a square of 150 to 200 kilometers per side.

Weather and climate are, obviously, far more localized than that. Mountain ranges–even individual peaks and valleys–make their own weather. A single glacier might grow or dissolve because of temperature and rainfall changes in a very specific area. Differences in air temperature over water and land cause breezes that can dramatically influence climate and weather in coastal areas. Regional models that take such phenomena into account are familiar to any viewer of TV weather news. But where global models are calibrated against data spanning decades, regional models are used to project only a few days ahead. Thus, one goal of climate scientists is to find a way for the twain to meet, to give local precision to predictions about global warming and climate change.

Rosenzweig’s group has calculated that today’s one-in-100-years New York flood would, in the 2080s, be considered a one-in-40-years or perhaps a one-in-four-years event. The order-of-magnitude difference is simply the result of variations between models. To calculate a more useful range of probabilities, Rosenzweig is currently combining global models with regional ones. By “nesting” models of smaller regional areas in the global grid boxes, she hopes to increase the resolution of climate-change predictions to 10 to 15 kilometers. In six-hour time increments, a global model introduces a fresh batch of climate variables into the regional models, which then make local calculations.

The project is ongoing; so far, efforts to validate such nested regional models against actual temperature measurements have shown the predictions to be off by 1 ºC or more, an unacceptable margin of error. Still, Rosenzweig expects that regional models will become more precise with further work. And as they do, one of their uses will be to better predict storm surges by accounting for changes in local wind patterns. “The large majority of climate impact studies have been done with the GCMs,” Rosenzweig says, referring to global climate models. “We are just now beginning to do more with the RCMs [regional climate models], and they are very much in research mode. Sea-level rise is the number one vulnerability, and we need better information for the agencies. It’s critical for their planning.”

To be sure, global sea-level projections are still a matter of debate: the IPCC pegged the 21st-century increase at between 18 and 38 centimeters under a scenario that assumed lower greenhouse-gas emissions and between 26 and 59 centimeters with higher emissions. This uncertainty makes perfect storm-surge predictions impossible. But the lack of information about local climate change remains a stumbling block that prevents New York City–and every other coastal area–from developing the detailed information it can act on. “You don’t always protect against the worst case, because you would bankrupt the city,” says Rohit Aggarwala, director of long-term planning and sustain­ability under New York’s mayor, Michael Bloomberg. “How urgent is it to invest in multibillion-dollar projects? Knowing that over the whole Atlantic seaboard there will be x sea-level change and x change in violent storms doesn’t necessarily help New York City make different decisions than Miami or Halifax.” On the other hand, he notes, if New York were to operate on incorrectly optimistic information and delay the most ambitious storm-surge barriers too long, the consequences could be disastrous.

New York City authorities have already gotten some specific warnings from Rosenzweig’s group, which made a study of how the city’s water-supply and sewage-treatment infrastructure could be affected by rising sea levels. For example, a pump station north of the city on the Hudson River–built to draw emergency fresh water during times of drought–will eventually require expensive new filtration systems as rising seas push salinated water to within range of the intake areas.

But while there’s still uncertainty about the rate at which sea levels are rising, it has become increasingly clear that temperature increases alone could severely tax a large city’s infrastructure. Late last year, the Union of Concerned Scientists in Cambridge, MA, released a report titled “Climate Change in the U.S. Northeast.” Produced in collaboration with climate scientists, the report predicts that by midcentury, northeastern cities could be experiencing an average of 30 to 60 days of temperatures above 90 ºF each year, up from 10 to 15 days historically. By the end of the century, these cities could see 14 to 28 days of temperatures over 100 ºF, if the higher-emission scenarios are realized.

Armed with such predictions, the city of New York and a prominent regional civic-planning group, the Regional Plan Association (RPA), are starting to think about how to respond. Jennifer Cox, a senior planner and director of geographical information systems at the RPA, is superimposing estimates of heat waves and storm surges onto city maps showing topography and socioeconomic characteristics. And GISS is collaborating with a consortium of universities whose members are now plugging temperature estimates into air-quality models, to see how bad ozone levels could get during the hotter summer days of the 2040s or 2060s. High ozone levels could produce severe health crises as heat waves overwhelm emergency facilities, water supplies, and the power grid.

But such studies are just the first academic pass at planning. The scenarios they envision are still relatively vague. And while suggested remedies abound, they reflect more imagination than engineering. Physical oceanographer ­Malcolm ­Bowman of the State University of New York at Stony Brook, for one, would place a tidal-surge barrier at the Verrazano Narrows (between Brooklyn and Staten Island); another near the Throgs Neck Bridge, where the East River meets Long Island Sound; another between Perth Amboy, NJ, and Staten Island; and a fourth across Rockaway Inlet at the entrance to Jamaica Bay. The barriers–more ambitious versions of the storm-surge barrier at the mouth of the Thames River outside London–could theoretically prevent tens of billions of dollars in damage. With the models and computational power available now, however, it’s hard to determine whether and when such ideas need to be acted on. “If you look at European experience,” says ­Bowman, “it takes a major flood and a major loss of life to get the bureaucracy to do anything.”

The Dry West

Colorado Springs, CO, is a boomtown in an arid region–just one of many cities that rely for water on the melting snowpack of the nearby mountains, delivered via the Colorado River and Arkansas River watersheds. Many other cities get their water similarly from the Sierra Nevada Mountains of California. But right now, the western United States is facing a slow-motion water-supply catastrophe wrought by climate changes that will inexorably reduce the snowpack. “The western U.S. is really not in good shape at this point,” says Linda Mearns, a climatologist at the National Center for Atmospheric Research (NCAR) in Boulder, CO, where she is director of the Institute for the Study of Society and Environment. This has become fairly clear “even without the regional detail” in climate models, she adds.

But the regional detail is still important for deciding how, where, and when to respond. Consider the Homestake Reservoir. High in the Rocky Mountains, not far from Vail, CO, it is part of a network of reservoirs and pipelines that feed water to Colorado Springs. In June 2006, the reservoir filled at the unprecedented average rate of nearly two feet per day. Because of higher temperatures earlier in the season, the snowpack was melting more quickly than usual.

The unprecedented may become routine as global warming makes more precipitation fall as rain, while what snow there is melts ever faster. That’s worrisome: a reservoir that fills more quickly than expected can stress a dry levee. And there are other concerns. At what point will earlier snowmelt translate into summer water shortages? Will early spring torrents raise the risk of downstream flooding? Will more-intense spring rainfalls increase sediment, overwhelming filtration systems and washing more pollutants into the water supply? And these climate-related questions arise at a time when rapid population growth is already stressing water resources.

Planners need to understand as precisely as possible the amount, timing, and form–rain or snow–of future precipitation. Only then can they determine when and where to build new water-storage, flood-control, and filtration systems and how to guide future residential or commercial development in watershed areas. So last winter, in a windowless conference room in an industrial area of Colorado Springs, engineers from Colorado Springs Utilities met with David Yates, an NCAR hydrologist, to start revising their water-supply management plans in light of climate-change projections. “Plans are typically made based on historical 20- to 40-year stream-flow averages,” Yates said. “That mode of planning is no longer relevant.”

Concrete local projections are especially important in this region, where politics constrain the way scientific findings can be discussed. Colorado Springs is a politically conservative city and home to a powerful Christian evangelical organization that is skeptical of concerns about global warming. Toward the end of the meeting, the utilities’ manager of water-supply resources, Wayne ­Vanderschuere, entered the room. He was already thinking about how any new climate-change-related findings might be framed. “All the talk about climate change and CO2–we don’t want to go there. We don’t want to talk about Kyoto, all the posturing,” he said, referring to the Kyoto Protocol, the U.S.-rejected treaty that mandates limits on greenhouse-gas emissions. “We just want the analytic risk to supplying water that this poses.”

Brett Gracely, the utilities’ planning supervisor, said Colo­rado Springs was at a turning point. “We’re trying to get a handle on what this all means for us,” he said. He–and his city, and the rest of the country–just aren’t sure exactly when and to what extent global trends will influence regional trends and make existing hydrology models obsolete. “If it comes down to literally building a model, how do we do that?” Yates asked during the meeting. “What needs to be done–what resources are needed to do that?” Colorado Springs’ effort to build a model of how climate will affect local hydrology has just begun and could take two years.

One strategy the model is likely to use is to break down mountainous regions into elevation bands rather than small, uniform grid boxes. Mountain areas, with their myriad microclimates, are particularly difficult to model. Mountains can cause winds to shift and clouds to form; snow-covered north faces, warm south faces, and cold valleys can give rise to strikingly different conditions. L. Ruby Leung and Steve Ghan, climate physicists at the Pacific Northwest National Laboratory in Richland, WA, are pioneers in the elevation-band approach, which they say can be as accurate as nesting techniques that zoom into smaller regions to resolve mountain effects, yet less computationally expensive. Their models provide, among other things, detailed pictures of how global warming will cause snow lines to move to higher altitudes, making it possible to estimate the resulting diminution of the snowpack that now provides most of the fresh water in the western United States. The state of California has already estimated that under scenarios assuming medium to high levels of future greenhouse-gas emissions, higher temperatures could eat away between 70 percent and 80 percent of the Sierra snowpack by century’s end.

But so far, the models Leung and her colleagues have developed are not reliable enough to dictate specific measures for addressing snowpack loss–like building new reservoirs. Water-resource managers want more certainty, Leung says, so she is pushing the research on several fronts by running multiple global and regional models. Such efforts are both labor and computation intensive, but they are also critical to reducing uncertainty. “We know that the climate is actually changing, but they are managing the water system based on rules devised 50 to 100 years ago,” Leung says. “If what we project in the future is correct, we expect a pretty big problem.”

Visualizing Water Shortages
Critical to understanding future water shortages in the western United States, the model that generated this image depicts springtime snowpack at a resolution of 1 kilometer, far better than the 150-­kilometer resolution of the average global climate model. Red peaks indicate the deepest snowpack; purple areas indicate none. (The vertical scale is exaggerated; California’s central valley is in the foreground.) This visualization incorporates data on clouds, surface temperatures, and precipitation, broken down by topographical elevation in a technology pioneered at Pacific Northwest National Laboratory. Zooming in on different regions reveals future snowpack loss on specific mountains (see “Vanishing Yosemite Snowpack,” p. 70).

The Canaries

The rich can buy state-of-the-art climate science, just as they can buy state-of-the-art health care. And upstream from Colorado Springs, the resort town of Aspen, CO, kicked off its very own climate-change impact study two years ago. City leaders called it the Canary Initiative–because in their view, mountain areas could be the climate-change equivalent of the canary in the coal mine. They sought out the advice of leading lights in climate science and devoted a modest $145,000 to a pair of local studies. The announcement of the studies’ findings last year bore the somber headline “Aspen Climate Study Finds Serious Risk to the Future of Skiing.”

At first blush, the emphasis on skiing may provoke eye rolling. But as Aspen goes, so goes any other mountain area. Aspen’s leaders have come to grips with the fact that by the end of the century, there may be too little snow not only for skiing but for replenishing water supplies, sustaining fishing, and fighting fires, which would themselves be more frequent in water-starved forests. “When I first heard about this, I thought it was surprising,” says Mearns, who participated in the Canary Initiative. “Little Aspen is going to do this full assessment? But as it went through, I realized this does make sense, because ultimately, mitigation on local scales can help.” Aspen may need to put ski lifts higher up the mountain and, eventually, plan for life after skiing.

If Aspen is, in its planning for climate change, an extreme exception at the local level, California is the exception at the state level; the California Energy Commission does more than any equivalent state agency to promote energy-efficiency technologies and renewable electricity sources like solar and wind. As a result, California’s carbon dioxide emissions from power generation are, per capita, the lowest in the nation. But in the past three years, the energy commission’s R&D effort has expanded to include studies of climate adaptation. A small portion of its annual climate budget, about $4 million, now goes toward regional computer modeling aimed at clarifying the implications of climate change for agriculture, forests, coastal management, and water supplies. “We want to know at higher resolution how the California snowpack and its water availability will be affected and, in turn, how agriculture and urban water-supply strategies may have to respond,” says Martha Krebs, the commission’s deputy director for R&D. “We need to know how sea-level rise and salt-water intrusion may affect coastal communities and what that will mean for city planning and development policy.”

Krebs anticipates that detailed regional models will, among other things, steer forest-management plans in new directions. California is considering the possibility that its forests could serve as “sinks” for carbon dioxide. One proposal is to reforest previously forested areas. But a hotter climate could stress existing plant species, and harsher droughts could leave these new forests more vulnerable to fire. The hope is that understanding the effects of climate change on California’s complex topography and climatic zones will help foresters develop the right management strategies, including choosing specific trees that can survive harsh conditions.

California–battered by population growth, landslides, and water shortages–is already taking action on flood-­control projects and considering how to protect, for example, fresh-water intakes in the Sacramento River delta, which is expected to become more brackish. Now it’s trying to respond specifically to climate change as well, but progress is slow. Coastal real-estate development policy, for example, hasn’t changed. “How well prepared is coastal California to deal with the impact of climate change?” asked Susanne Moser, a coastal-zone expert at NCAR who is advising the state. “The bottom line is, there are very, very few counties and municipalities that are doing anything about this topic so far.”

Staggering Backwards

On a wall-size screen inside the darkened Visualization ­Laboratory at NCAR, the display of global temperatures across two centuries gets rolling. The animation, which uses a middle-range estimate of future greenhouse-gas emissions, starts in 1870; small splotches of blue and yellow–minor temperature deviations from historical averages–flash over a slowly spinning earth. In the mid-1880s, more blues appear; the planet cooled for a time, thanks to atmospheric dust kicked up by the massive Krakatoa volcanic eruption in August 1883. Things level out at the turn of the century and remain steady through World Wars I and II.

But from the 1950s through the 1980s, yellow blotches proliferate. In the 1990s, the jaundice spreads; by 2005, the earth looks disturbingly like a glowing yellow tennis ball. By the 2050s, the top of the planet appears red. And by 2099, much of the world has been painted orange and red by global warming.

Another model shows the projected changes in seasonal expansion and contraction of Arctic sea ice as the years roll by. It’s like a slowly diminishing heartbeat, with summer ice gradually vanishing.

Caspar Ammann, an NCAR scientist, offers some perspective on the disturbing show. A temperature increase at the upper end of the IPCC’s projections–5 ºC by the end of the century–is about the same size as the increase that’s occurred since the depths of the last ice age. In other words, during the ice age, it was 5 ºC cooler than it is today. If global temperatures rise 3 ºC above recent averages, they will be in the vicinity of temperatures last seen three million years ago, when sea levels were at least 15 meters higher–though it could take centuries for ice sheets to melt and raise the oceans that much. “This is the magnitude of the [temperature] change that is possible in 100 years,” says Ammann. “We need to see that perspective clearly.”

Two days after I saw the NCAR simulations, I visited Ted Scambos, lead scientist at the University of Colorado’s National Snow and Ice Data Center (NSIDC) in Boulder. Scambos studies ice dynamics to understand the rate at which the ice sheets of Antarctica and Greenland are responding to climate change. He and other scientists at NSIDC spend their days poring over satellite data, studying how glaciers slide down ancient hidden fjords and how warmer ocean water and the glaciers’ own meltwater lubricate their progress. “We are warming so fast that the earth is still staggering backwards from the warming,” Scambos said. “We may have already crossed the threshold of the last warm period, a time when people were growing grain in Iceland and raising dairy cattle in southeastern Greenland. And even if you flattened out greenhouse emissions right now, my hunch is that all the arctic sea ice in summer will eventually disappear.

Vanishing Yosemite Snowpack
These images–which show details of the one on page 65–reveal snowpack changes in a 100-kilometer swath of California’s ­Yosemite National Park. The top image shows current spring conditions, with snowpack depth depicted by colors (greener indicates less snow, whiter indicates more). The bottom image is a visualization of snowpack between 2050 and 2070. Contour lines depict percentages of snowpack lost, which range from 80 percent (light blue contour) near today’s snow line to 20 percent (red contour) near peaks. This level of topographic precision is critical to helping communities plan for changes in the amount of annual snowpack and the timing of melting, on which millions of people depend for their water supplies.

“We’re really, really in trouble,” he continued. “It’s just a question of time. People say climate has changed before and people adapted. That is true. But there weren’t six billion of us, with all the arable land working as hard as it could, and every one of those areas counting on climate more or less staying the same. All our infrastructure is built around this climate. Personally, I think we have a strong moral obligation to respond in a fashion that gives people a century from now a reasonable chance of making their way ahead. We should do something.”

The ability to “do something,” however, depends on getting information that is much better and more detailed. And that will depend on increasingly precise computer models and more monitoring equipment to feed data into those models. Not every city has a Goddard Institute for Space Studies in its backyard, Cynthia Rosenzweig points out. She says every local government should be given the tools to understand how global warming will affect its community. “We need a national capacity for scenarios, to provide every locality in the nation with the input variables they need for projecting impacts and preparing adaptations,” she says. “We should begin to incorporate sea-level rise into plans for coastal development. We should improve our responses to heat waves–now–so we can be prepared for greater frequency and duration. And we should consider the potential for more droughts–how we would manage for more droughts and floods.”

But from NASA to the NOAA to the National Science Foundation and the U.S. Department of Energy, the budget picture is dismal. In 2005 dollars, the annual federal budget for climate-change research has been slashed from more than $2 billion in the mid-1990s to less than $1.6 billion today. Earlier this year, a National Academy of Sciences report warned that Earth-observing satellites–basic hardware for monitoring climate change–were at “great risk” of blinking out. Without urgent investment, the report warned, 40 percent of sensors and other instruments aboard NASA spacecraft could stop functioning before the end of the decade. “At these agencies, earth-science and climate­-­science budgets are either level or decreasing in real dollars,” says MIT’s Ronald Prinn. “Under those circumstances, what is needed for helping out states and cities is just not going to appear. It is a sad state of affairs. At a time when we should be trying to help at the regional to the local level, with sound advice, we are facing this incapa­bility to have accurate forecasts at the local level that make the advice worth taking.”

David Talbot is Technology Review’s chief correspondent.

Keep Reading

Most Popular

This new data poisoning tool lets artists fight back against generative AI

The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models. 

The Biggest Questions: What is death?

New neuroscience is challenging our understanding of the dying process—bringing opportunities for the living.

Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist

An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.

How to fix the internet

If we want online discourse to improve, we need to move beyond the big platforms.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at with a list of newsletters you’d like to receive.