Peak Water: Aquifers and Rivers Are Running Dry. How Three Regions Are Coping

That the news is familiar makes it no less alarming: 1.1 billion people, about one-sixth of the world's population, lack access to safe drinking water. Aquifers under Beijing, Delhi, Bangkok, and dozens of other rapidly growing urban areas are drying up. The rivers Ganges, Jordan, Nile, and Yangtze — all dwindle to a trickle for much of the year. In the former Soviet Union, the Aral Sea has shrunk to a quarter of its former size, leaving behind a salt-crusted waste.

Water has been a serious issue in the developing world for so long that dire reports of shortages in Cairo or Karachi barely register. But the scarcity of freshwater is no longer a problem restricted to poor countries. Shortages are reaching crisis proportions in even the most highly developed regions, and they're quickly becoming commonplace in our own backyard, from the bleached-white bathtub ring around the Southwest's half-empty Lake Mead to the parched state of Georgia, where the governor prays for rain. Crops are collapsing, groundwater is disappearing, rivers are failing to reach the sea. Call it peak water, the point at which the renewable supply is forever outstripped by unquenchable demand.

This is not to say the world is running out of water. The same amount exists on Earth today as millions of years ago — roughly 360 quintillion gallons. It evaporates, coalesces in clouds, falls as rain, seeps into the earth, and emerges in springs to feed rivers and lakes, an endless hydrologic cycle ordained by immutable laws of chemistry. But 97 percent of it is in the oceans, where it's useless unless the salt can be removed — a process that consumes enormous quantities of energy. Water fit for drinking, irrigation, husbandry, and other human uses can't always be found where people need it, and it's heavy and expensive to transport. Like oil, water is not equitably distributed or respectful of political boundaries; about 50 percent of the world's freshwater lies in a half-dozen lucky countries.

Freshwater is the ultimate renewable resource, but humanity is extracting and polluting it faster than it can be replenished. Rampant economic growth — more homes, more businesses, more water-intensive products and processes, a rising standard of living — has simply outstripped the ready supply, especially in historically dry regions. Compounding the problem, the hydrologic cycle is growing less predictable as climate change alters established temperature patterns around the globe.

One barrier to better management of water resources is simply lack of data — where the water is, where it's going, how much is being used and for what purposes, how much might be saved by doing things differently. In this way, the water problem is largely an information problem. The information we can assemble has a huge bearing on how we cope with a world at peak water.

That data already shows the era of easy water is ending. Even economically advanced regions face unavoidable pressures — on their industrial output, the quality of life in their cities, their food supply. Wired visited three such areas: the American Southwest, southeastern England, and southeastern Australia. The difficulties these places face today are harbingers of the dawning era of peak water, and their struggles to find solutions offer a glimpse of the challenge ahead.

On the descent into Sky Harbor International Airport, Phoenix's endless grid of streets and tract homes is etched into the desert floor like the imprinted surface of a microchip. When the sunlight hits at the right angle, the canals that zigzag across the landscape light up like semiconductor traces surging with electricity.

And Phoenix is sprawling at a rate that seems to rival Moore's law. In the 1990s, the metro area was growing at the rate of an acre every three hours. The population is expected to nearly double in the next 20 years. But cities, unlike microchips, don't double in efficiency every 18 months. A 2007 government report stated that staggering growth in the American Southwest "will inevitably result in increasingly costly, controversial, and unavoidable trade-off choices." The issue: how to parcel out a dwindling water supply.

The city's chief water sources are the Salt River Project and the Central Arizona Project, two massive water systems that bookend a century-long effort to hydrate the region. The Salt River Project began in 1903 with the Roosevelt Dam, which reined in the flood-prone waterway. Today, the SRP is a vast network of reservoirs, hydroelectric dams, and channels. As for the Central Arizona Project, it's one of the largest and most expensive aqueducts in the US, completed in 1993 at a cost of $3.6 billion. The 336-mile CAP canal diverts 489 billion gallons a year from the Colorado River, irrigating more than 300,000 acres of farmland and slaking the thirst of Phoenix and Tucson.

 

The CAP isn't the only straw sucking at the Colorado. Seven states and dozens of Indian reservations, as well as Mexico, tap its flow. Development has sapped the river, a problem exacerbated by a drought called "perhaps the worst in 500 years" by US interior secretary Gale Norton. Lake Mead, an immense reservoir that dams the Colorado to supply most of Phoenix's water, has a 50-50 chance of running dry by 2021, according to a study by the Scripps Institution of Oceanography. Larry Dozier, the CAP's deputy general manager, calls this finding "absurd," claiming that studies show the reservoir won't disappear entirely, even in the worst case. However, the Scripps researchers counter that their calculations are conservative and warn that "the water shortage is likely to be more dire in reality."

Chandler, a city on the southeastern edge of Phoenix, epitomizes the regional dilemma. Founded in 1912 to accommodate farmers who ventured into the Sonoran Desert, Chandler supports a population that has tripled in the past 20 years to 250,000. On the outskirts of town, where the last remaining farms fade into the scrub, stand three colossal Intel semiconductor manufacturing plants: Fab 12, Fab 22, and the gleaming new Fab 32, which produces state-of-the-art chips on a floor area equivalent to 17 football fields. Intel is a key driver of the local economy. The company employs 10,000 people and has invested $9 billion in Chandler; its workers, on average, earn four times the Arizona median salary. Just one problem: The fabs are also by far the city's biggest consumer of water.

Chip fabrication is a thirsty process. The silicon wafers must be rinsed after each of the several dozen semiconductor layers is applied and etched. Consequently, the Intel campus has been designed to maximize every drop of the 2 million gallons it uses daily. Intel, wary of spilling its manufacturing secrets, bars journalists from entering the enormous silver and white monolith. Fortunately, the plant's circulatory system is visible from the outside. Len Drago, who is responsible for the facility's environmental profile, offers to show it to me. As we walk around the building's perimeter, he explains how water flows through the plant.

The tiniest imperfection can render a wafer useless, so incoming water cascades through a series of filters until its mineral content is a hundred-thousandth that of Colorado River water. The briny byproduct goes into a towering tank that looks like a Jules Verne moon rocket, which distills out the remaining water and pumps it back to the beginning of the system. The salty sludge goes to an evaporation pond. The purified water, meanwhile, is used to wash chips. The rinse water is treated and then sent to other parts of the campus: the air scrubbers that filter the plant's emissions, the massive cooling towers that keep workers from suffocating in the desert heat. Even the drought-resistant desert landscaping in the plant's parking lot is irrigated with wastewater.

But Intel doesn't reuse all of its wastewater. Every day, the company pumps 1.5 million gallons to a $19 million reverse-osmosis desalination plant it built for Chandler. This water, cleaned to drinking standards, is pumped 6 miles away and injected 600 feet down into a sandstone aquifer beneath the city. To date, Intel has banked more than 3 billion gallons. The facility recycles or stores 75 percent of the water it brings in, Drago says.

Intel isn't simply trying to be a good corporate citizen. Nor is it merely out to save money. Running a sustainable operation greases the regulatory wheels when the company wants to expand. Because Intel was well within the government's environmental thresholds for the site, Fab 32 didn't even require a new water-use permit. It hasn't always been this way, Drago admits. "Back in the early 1980s, we had three Superfund sites in California," he says. "It's a lot easier to do things the right way. Especially in the long term."

The long term, however, will be ruled by the twin realities of an exploding population and a hotter, drier climate. Dave Siegel, Chandler's water czar, describes how he plans to continue providing for the growing city (and his biggest customer, Intel). The government has legal rights to all the water it needs, he says, not only from the SRP and the CAP but from 27 wells drilled into the aquifer. "That's legal water, mind you," he says. "It's a different thing than physical water." Legal water refers to the complex array of agreements, treaties, and laws that govern water use in the American West — and federal and state allocations trump Chandler's municipal rights. As for physical water, that's the stuff coming out of the tap. All the legal water in the world isn't enough to wash a bandanna if there's no physical water available.

So Chandler came up with a clever plan. The city banks as much excess CAP water as it can, pumping it underground along with Intel's contribution. Thanks to this so-called recharge, the local aquifer is actually rising a few feet a year. Siegel maintains that even if the most apocalyptic predictions came true — say, the rivers collapse completely — Chandler would be able to soldier on. "If we never recharge another drop," he says, "we have enough water underneath us to last about 100 years." His projection includes future growth, including two more Intel fabs now on the drawing board.

 

But many scientists say that banking river water underground is not enough. Gary Woodard, a bearded water resource expert at the Sahra Center at the University of Arizona in Tucson, has made a career of studying the issues facing water-stressed regions around the world. He admires Intel's efforts, but he cautions that direct water consumption is only half the story. To describe the other half, he invokes the "water-energy nexus": the idea that it takes water to produce energy, and energy to take advantage of water. That is, supplies of water and power are interdependent.

"Intel is doing everything it can," Woodard says, "but high-quality recycling, pumping water up and down and recirculating it, uses an incredible amount of energy." The Intel campus draws the power equivalent of 54,000 homes every year. Intel gets a sizable portion of that power from the Palo Verde Nuclear plant outside of Phoenix, and this means that it takes far more water to make a microchip than actually circulates through the company's recycling system. "No energy generation system uses more water than a low-desert nuclear plant," Woodard says. Palo Verde uses 20 billion gallons annually to cool its turbines. That water is emitted as water vapor from its cooling towers, to fall as rain somewhere else. None of this is figured into Intel's water footprint. Neither are the additional employees, new homes built in the desert, and cars that will come along with Intel's next round of fabs. More parking lots will absorb more solar radiation, contributing to Phoenix's urban heat island. More energy and more water will be required for cooling.

Other experts share Woodard's concern. Peter Gleick is president of the Pacific Institute in Oakland, California, a leading think tank on water issues. He isn't surprised that Intel and Chandler are optimistic about the future. Their cheerful attitude, he believes, reflects their confidence that social and economic priorities are on their side. "It shows to what lengths we'll go to ensure water for high-value uses," Gleick says. "Truth is, Intel will always be able to pay more than anybody else for water. They can act as though it's not scarce, because for them it's a relatively small cost."

If moneyed special interests determine the going price of water, eventually they will edge out users who can't afford to pay top dollar. Agriculture will be squeezed out, as will water rights for poorer communities. And the environment, it goes almost without saying, will twist in the wind: "That's why the Colorado River no longer reaches its delta," Gleick says. Intel will continue to drive Chandler's economy, lubricate the regulatory process, and burnish its image while small towns elsewhere along the Colorado wither. If this seems mercilessly Darwinian, it's also true that Intel has a critical role to play in solving water issues. Ever-more-capable microprocessors are at the heart of efforts worldwide to keep the water flowing.

Looking out at Kensington Gardens in London, where ornate fountains shimmer in the sunshine, it's difficult to imagine that this famously damp city has less water per person at its disposal than Dallas, Rome, or Istanbul. But it's true, and the problem is getting worse. I'm sitting in a restaurant next to the gardens with John Rodda, a hydrologist with Britain's Centre for Ecology and Hydrology. White-haired and buttoned-down, Rodda presents his doom-and-gloom outlook with quintessential British stoicism. He pulls out a map of Britain and points to the country's southeast, printed an angry shade of red to indicate water scarcity. "We fall well below the World Bank standard, per capita, of a water-stressed region," he says.

In the summer of 2006, London was hit by the worst drought in three decades. After two consecutive dry winters (the time of year when rainfall usually replenishes the water supply), the city imposed restrictions on watering lawns, filling swimming pools, and other nonessential uses. Newspaper columnists raised the specter of Londoners lining up at fire hydrants to collect water rations. Desperate to maintain supplies, water companies considered extreme measures: cloud seeding, bulk transportation by tanker, even towing icebergs down from the Arctic.

Unlike Arizona, where industry and agriculture use the vast majority of the water, London serves mainly the people who live there. But there are a lot of them: 7.5 million, expected to exceed 8 million by 2016. "We've got a huge number of people living on a small island where it doesn't rain as much as people think," says Jacob Tompkins, director of Waterwise, a London-based nonprofit devoted to water efficiency, "and we're living in the driest bits."

The 2006 drought made it clear that anything more severe, like a longer run of dry winters, would push the system toward collapse. "It would shut down the economy," Tompkins says. Then there's the changing pattern of rainfall. After the 2006 drought, the summer of 2007 was one of the wettest on record. But that rain fell in downpours rather than the customary drizzle, causing devastating floods. "It used to rain the same amount every year," Tompkins continues. "We built some reservoirs and it was fine. But rainfall intensity has doubled, and rain comes in storm events. In terms of infrastructure, that doesn't work very well."

 

London's infrastructure has a more fundamental problem: It's creaking with age. "Charles Dickens was the best-selling author when most of our pipe work went in," says John Halsall, director of water services at Thames Water, the private company that provides water to greater London. Thames Water maintains more than 300 reservoirs, 99 treatment plants, and more than 20,000 miles of pipe. The city's water system was a triumph of 19th-century engineering, but one-third of the mains are more than 150 years old, veterans of such scourges as Hitler's bombs and corrosively acidic soil. Thames' system leaks 180 million gallons a day, 30 percent of overall flow. To fix a leak, which the company does some 82,000 times a year, it has to shut down traffic and dig up the streets in one of the most congested cities on Earth. A brief walk around the West End turns up a half-dozen work crews digging up Victorian mains, scooping through layers of history to repair the pipes one segment at a time.

Replacing all the Victorian pipes would cost an estimated $3.6 billion. The conundrum facing Thames Water is how to upgrade the crumbling system without tearing up the city or bankrupting the company. There are two sets of solutions: On one hand are small, local, high tech projects. On the other are traditional large-scale civil engineering initiatives that have been a staple of water management since the Roman Empire. Tompkins favors the small-scale approach. In particular, he likes metering. There's no way to measure the water flowing through much of the underground infrastructure, which makes it hard to identify leaky sections. Likewise, not even a quarter of the city's households are metered, and that makes it difficult to encourage conservation. If consumers understood exactly how much they were using, Tompkins reasons, perhaps they would change their behavior, like a dieter motivated by the scale readout every morning.

Metering addresses a lack of information about water at the lowest level, right in the pipe. And now there's a way to do it more effectively than ever. In the grand tea room of a posh business club known as the Institute of Directors, overlooked by huge oil portraits of admirals and lords, Michael Tapia shows me a device called iStaq. Tapia is CEO of Qonnectis, iStaq's manufacturer. Barely the size of a hardcover book, the unit can be tucked away under a manhole cover and transmit measurements of water level, pressure, flow, and other variables. "The system itself is intelligent," Tapia says. "It will send you an email or text saying, You have a burst pipe.'" Qonnectis has a $400,000 contract with Thames Water to help detect leaks.

Electric, gas, and water utilities all stand to benefit from smart metering. So far, smart water meters have been deployed mostly in oil-rich Middle Eastern cities like Doha and Abu Dhabi, where water is precious and infrastructure relatively new. However, simply measuring the flow is a surprisingly powerful motivator. Research shows that installing a meter in a house so people can see how much water they're using can reduce consumption by 10 percent. With the right political carrots and sticks, Tompkins estimates, 70 percent of the city's homes could be metered in just over a decade. "People need to get away from the idea that you just turn on the tap and all the water you want is there," he says.

As promising as smart meters are, giant utilities like to think big, and to them metering is only one drop in an Olympic-size pool. Thames Water has grander designs. The company hopes to dig a 20-mile drainage tunnel, called the Thames Tideway, underneath the river to its sewage treatment plant. The structure would be a hedge against climate change, designed to prevent the city's sewers from flushing into the watercourse as storms intensified. And to address storage capacity, plans call for a huge new reservoir in Oxfordshire. But projects of this scale can take 20 years to complete, and the company is under pressure to find new supplies sooner than that.

Thames Water's most controversial project is a $400 million desalination plant called the Thames Gateway. The proposed facility could take in seawater, filter out the salt, and deliver 35 million gallons of drinking water a day during drought emergencies. Desalination would essentially drought-proof the city, the company claims. It's an appealing solution. The ocean is practically limitless, and the plant would run on biodiesel, giving it a green imprimatur. The project was moving through the approval process in 2006 when London's tough, left-leaning mayor, Ken Livingstone, blocked it.

Livingstone argued that the plant was too expensive and that desalination is too energy-intensive. Stripping seawater of its salt is a pricey way to obtain freshwater, cost-effective only for high-end uses like drinking, but not bathing or watering gardens. And the mayor questioned the proposal's environmental cred: Biodiesel emits carbon, and desalination's super-salty byproduct is toxic to marine life. Thames Water would do better, he insisted, to repair London's decrepit labyrinth of pipes.

With the desalination plant deadlocked, London is running out of time. "The big projects we do are taking longer and longer to get approval, and it doesn't take much to throw them off track," Halsall says. "While we're debating, the risk increases to our basic supply."

Australia has always been dry. It's the most arid continent after Antarctica. Covering an area roughly the size of the lower 48 states, it supports less than one-tenth the US population, and even that is an enormous strain on water supplies. The country was founded during the second-worst drought in its history, but the worst dry spell is unfolding right now. Rainfall, which has declined to 25 percent of the long-term average, is projected to plummet another 40 percent by 2050.

Three factors are working to desiccate the landscape. One is simple overexploitation of existing resources. More water is withdrawn to support agriculture, industry, and cities than the system can handle. Another is El Niño, a weather pattern that periodically alters rainfall, further drying the continent. The third is climate change. Australia is growing hotter, which compounds the other two problems by boosting both consumption and evaporation.

The convergence of these factors could have catastrophic results. Every major city in Australia is hobbled by mandatory restrictions on water consumption, but most of the country's water — two-thirds — goes to agriculture. The economics of food production have always been based on ready access to cheap water. The price of beer has been rising since a jump in barley prices, a development that many joke could lead to large-scale civil unrest. But it's no joke: The global price of wheat hit its highest level in decades in December, partly due to Australia's water shortage. The most fundamental impact of scarcity will be on Australia's ability to feed itself.

Two hundred miles north of Melbourne, in a dusty farmyard in Moulamein, New South Wales, L. J. Arthur slides open a large steel barn door and steps into the shadows. A few minutes later, the silver-haired, 53-year-old rice grower emerges pushing a helicopter on detachable wheels, the tail rotor braced against his shoulder. We clamber into the bubble cockpit. "From the air, you'll get a much better sense of what two years with no water looks like," he says, checking the gauges. A cloud of dust billows around us, and the pitch of the rotor's whine rises as we pull away from the ground.

We climb to 1,000 feet, and Arthur shouts over the engine, "In a normal year, this would be a carpet of every shade of green imaginable, rice fields as far as you can see." The landscape is tinder-dry, the fields a shorn patchwork of grays and browns, stretching in mind-boggling flatness toward the ochre wastes of the outback. "In a normal year, we'd have 1.2 million tons of rice under production. This year we have 15,000, and there's no telling if that will make it." Rice is often dismissed as the wrong crop for the region because it requires flood irrigation. But producers in the basin can grow 10 tons per hectare, among the highest yields in the world.

We swoop low over a stubble field, scattering a flock of emus. A huge red kangaroo lies languidly in the shade of a eucalyptus. The local native animals are doing fine for now, Arthur tells me, helped along by a watering pond he dug for his sheep. But the future of people in this arid corner of an arid continent is far less certain.

For the second year in a row, the region's rice farmers have received no water at all from the Murray River, the 1,500-mile lifeline that flows out of the Snowy Mountains and helps hydrate the cropland for 40 percent of Australia's food. Inflows into the Murray River last year were the lowest in 116 years of recorded levels, almost half the previous low. Reservoirs in the southern basin hold only 20 percent of capacity, and the summer drawdown hasn't begun.

No one who tries to eke out a living from this land is untouched. The 400,000-square-mile Murray-Darling basin, named for the two main rivers that run through it, receives only 6 percent of the continent's increasingly scarce rainfall. In some places, the groundwater is too salty to drink. Coastal cities are investing in desalination plants, but desalting technology is simply too expensive to use for agriculture. Without irrigation from the river, agriculture couldn't exist here. The farms would literally dry up and blow away.

We land near downtown Moulamein, where a dozen tractors are parked around a shallow clay pit the size of a shopping mall. Since no crops are in the ground, the government has hired local growers to dig the giant hole as an emergency reservoir for the town. Some cash-strapped farmers are now members of road-building crews. A nearby rice-processing plant laid off 90 workers, and the press has reported on depression and suicide among ruined farmers. Many small towns in the basin are teetering on the brink of economic collapse.

A few hours away from Arthur's farm, the managers of the Coleambally Irrigation Cooperative have set out to make Australian agriculture viable. The co-op is a group of 320 farmers connected by a 300-mile network of irrigation channels. Their section of the basin received just 3 percent of its water allocation in December. That means they'll have to become vastly more efficient. Bringing that about is the goal of Murray Smith, Coleambally Irrigation's CEO. In Australia, one-third of agricultural water, on average, is lost to leakage, seepage, evaporation, and faulty metering. Smith thinks the future of farming in Australia is "more crop per drop." Toward that end, his company has invested $15 million in a host of technologies to minimize waste.

In a back room at Coleambally headquarters, Smith calls up a series of displays on a computer screen showing real-time measurements of flow, temperature, and salinity at remote-controlled irrigation gates spread over thousands of acres. Software helps determine exactly where water is being wasted; problems can be addressed by opening or closing gates. Out where the canals draw water from the river, automated flume gates control the inflow. This kind of centralized management is revolutionizing irrigation. It's the same sort of network that allows engineers at Thames Water to watch over London's water supply, the same technology that lets Intel managers optimize the flow of millions of gallons through Fab 32.

There are other datastreams as well. In much the way an MRI depicts the inner workings of the human body, Smith's co-op is using electromagnetic imaging to map the hidden hydrography beneath fields, showing where buried streambeds lurk to draw away precious irrigation water. Sensors dragged through canals can help spot seepage, and sensors embedded in soil can help tailor the irrigation to a particular crop. Eventually, all of this data will be monitored by the farmers themselves through a single Web site, providing a more precise picture of water use than growers have ever had.

Not every farmer wants to risk being an early adopter. When Smith took over the company four years ago, some of the new systems were having problems. Gates didn't work, metering was off, and some crops were lost. Farmers were angry. Smith received death threats. "We're talking about people's livelihoods," he says. Still, Smith has faith in the co-op's network. "Nobody has ever integrated all these technologies into a single irrigation district. Coleambally is going to be the best in the world."

But what about the inescapable facts of drought, climate change, overuse, and scarcity? Smith acknowledges that pain is inevitable, and he envisions a fierce competition among the basin's farmers. Some farmers will be ruined. Some will cash out, availing themselves of one of Australia's newest innovations, an open market for water rights, where 1 megaliter (264,000 gallons) goes for $360. Those who survive will be the ones who use water most efficiently by planting less-thirsty crops and adopting better methods, and they'll have the market to themselves. "There are benefits to being the last man standing," Smith says.

The flow of water through a swath of drought-stricken farmland is complicated. The hydrology of an entire continent is mind-boggling. A day's drive to the east, in the leafy, ironically drizzly capital city of Canberra, I meet Stuart Minchin, a specialist in water information systems who works for the Commonwealth Scientific and Industrial Research Organisation. CSIRO's campus is spread across a lush, eucalyptus-covered hillside above the capital, and Minchin has invited me to see his baby, the Water Resources Observation Network. The centerpiece of the facility is a new $1 million computing and visualization center. I follow Minchin into a large, windowless space outfitted like a cross between the set of The Situation Room and the deck of a Star Destroyer. One wall is covered with several theater-size screens. A bank of computer monitors flashes graphics.

"The problem with both water science and water policy is that there's a vast amount of data and no easy way to understand it," Minchin says. "We're thinking about how to create spatial understanding of water issues." I don a pair of 3-D glasses, and a huge map of Australia leaps off one of the screens, scattered with dozens of blue bar graphs that seem to reach out toward me.

Information about reservoir levels in Murray-Darling used to be spread among 40 Australian agencies. So WRON set up a Web robot to screen-scrape the data and display it on a satellite map. A graphical slider tracks the levels by date. "It's a very powerful way to show information," Minchin says. He imagines using a Google Street View-type technology to map an entire watershed, down to the last wilting gum tree.

In an adjoining room, a huge bank of servers whirs: The Intel processors, whose manufacture is draining the Colorado River half a world away, are being harnessed to solve Australia's water crisis. "We've never known the answers to basic questions like how much water is in the entire basin," Minchin says. For more than a year, this supercomputer has been crunching 40 terabytes of remote sensing data. When it's finished this year, the analysis will shed light on the way water moves through the region and the consequences of human exploitation. It could hold the secret to restoring the Murray-Darling basin to health.

Minchin is confident that WRON will make a crucial difference to Australia's future — but he doesn't underestimate the challenge. "We'll never have a secure continent," he says. "But at least we can know what the limits are and try not to exceed them."

On the other side of the Snowy Mountains, Hume Dam corrals the Murray River into one of the largest reservoirs in Australia. When construction was finished in 1936, Hume was among the world's greatest public-works projects. It can store 400 trillion gallons and release them at will, providing a stable source of water to the farms and towns along the Murray River and securing Australia's economy for the foreseeable future.

In the sweltering heat, I walk across the concrete expanse of the spillway, more than a mile from end to end. Far below, Lake Hume is not even a quarter full.

In Peter Gleick's view, we have to move away from the "hard path," the massive civil-engineering projects and exploitation of untapped sources that defined the 20th century. Instead, we must turn to a "soft path," making the most efficient use of what we already have. Technology can help, and some new infrastructure will be necessary, Gleick believes. But the larger issue is conceptual: We must view efficiency itself as a source of water and tap this hidden wellspring. Americans already use 20 percent less water per capita than they did a generation ago. Gains in industrial use are even more impressive: A ton of US steel manufactured today requires just 2 percent of the water it did in the 1940s. Still, we are using more than we have. Can we change enough, and soon enough? "The whole point of peak water," Gleick says, "is that we have to fundamentally rethink who gets to use water for what."

It's the first day of Australia's summer, the beginning of Lake Hume's annual drawdown. Measuring rods poke out of the earth high above the lake's surface. Reaching far up the valley, a forest of dead gum trees, drowned decades ago when the lake was filled, is reemerging. Black spires poke their skeletal branches skyward. It's an eerie sight. In the worst-case scenario, Lake Hume will contract to 1 percent capacity this summer. The bed of the Murray River will be all that remains.

Matthew Power (matthewpower.net@gmail.com) has written for Harper's, Men's Journal, and The New York Times.