Puffy clouds contain a breathtaking amount of water. The volume of even a small one can top 750 cubic kilometers, and if you figure a half gram of water per cubic meter, those wispy balls of atmospheric fluff start looking like flying lakes.*

Now imagine you are a farmer watching them glide over drought-parched fields, carrying more than enough water to save your crops and pull you out of debt yet yielding only a few tantalizing drops before disappearing over the horizon. It is that maddening condition that leads people worldwide to spend millions of dollars every year trying to control the rain.

In the U.S., the desire to wring more moisture from the sky is growing particularly intense in this fourth year of severe drought. Over much of the Great Plains and the Southwest, rainfall since 2010 has been off by anywhere from a third to two thirds, with corn, wheat and soybean prices jumping by as much as a quarter. California, the source of much of the nation's fruits and vegetables, has yet to emerge from a three-year drought that has left reservoirs half-dry and snowpack dangerously low. In February, the National Weather Service gave the state a one-in-1,000 chance of recovering anytime soon. Almond farmers were bulldozing their trees for want of moisture, and even drinking water was threatened.

Worldwide, millions people are living in extreme drought, with 168 countries undergoing some level of desertification. Australia recently spent nine years in a drought they named the Big Dry. Turkey has been experiencing its worst drought in a decade. Brazil, China, and countries throughout the Middle East and South Asia have all recently faced drastic water shortages. And if the United Nations' World Meteorological Organization is correct, climate change is going to make things worse. Although only about 0.04 percent of the world's freshwater is floating in the atmosphere at any one time, it is the water we can get our hands on—if we are lucky. Or smart.

A few visionaries are experimenting with zapping the atmosphere with ions to squeeze more moisture out it, but the primary method of increasing rainfall is infusing, or “seeding,” clouds with chemicals. In 2012 nationwide a dozen operators in nine states ran cloud-seeding operations over more than 83,000 square miles. The Chinese government, for its part, deploys a “weather army” of 48,000 people armed with 50 airplanes, 7,000 rocket launchers and 7,000 cannons to coax more rain from the heavens.

The principle is simple. Clouds that could produce rain contain micron-size droplets of water whose temperature is below freezing but that have not yet turned to ice because they lack nuclei around which to form—say, dust particles of precisely the right size. The droplets are too light to counter the updrafts keeping them aloft. Provide suitable nuclei, though, and the droplets coalesce into pellets of ice. As they fall through the warm atmosphere, they turn to nourishing rain. Bernard Vonnegut, an atmospheric scientist at the General Electric Research Laboratory in Schenectady, N.Y., invented the technique in 1946, shortly after his little brother, Kurt, was released from the German POW camp that he would later immortalize in the novel Slaughterhouse-Five.

The chemical that Vonnegut used to seed clouds was silver iodide, whose molecular structure mimics that of ice crystals. In a cold cloud, it tricks the water into sticking to it. Silver iodide works in theory, and it even works in practice; pilots say they can see clouds change as the chemical hits them. But the question that has hung over cloud seeding for half a century is: Would that seeded cloud have rained anyway? There is no way to run a perfect controlled experiment. These are clouds we are talking about, and while a regiment of scientists today call themselves “cloud physicists,” clouds are the very definition of ephemeral. Each one is as unique as a snowflake and as skittish as a flame.

What we know about clouds is dwarfed by what we do not. Predicting what they will do is hard enough. Determining with certainty what they might have done under different circumstances is impossible. As recently as 2003, the National Research Council was skeptical. “There is ample evidence that ‘seeding’ a cloud with a chemical agent … can modify the cloud's development and precipitation,” read a summary of one of its reports. “However, scientists are still unable to confirm that these induced changes result in verifiable, repeatable changes in rainfall, hail fall, and snowfall on the ground.”

In the decade since the National Research Council's report, though, a fleet of new NASA weather satellites, advances in radar and the exponential growth of computing power have combined to let scientists say with considerable certainty—for the first time—that, yes, under the right circumstances and in limited ways, cloud seeding works.

Accumulating Evidence
“Water is an emotional thing. Drought is emotional,” Roelof Bruintjes said when I visited him at his sparsely furnished lair in one of the austere modern buildings that houses the National Center for Atmospheric Research (NCAR). Bruintjes, a courtly, elegant, Dutch-born cloud physicist, has been studying weather modification for decades and is only now starting to feel confident about its usefulness. “Farmers look up and see all this water passing overhead, and their fields are drying out, and they want the government to get them that water.”

It is hard to think of another scientific endeavor in which the fundamental technology has not changed in 70 years. Most cloud seeders are still putting plain-old silver iodide in clouds. What has changed, especially in the past 10 years, is the technology for evaluating its efficacy. In the 1980s came Doppler radar, particularly a version called 88D, which allowed scientists to see concentrations of water inside a cloud for the first time. This is the machine that throws those green splotches across the weather maps we see on television. “But even that is imprecise,” Bruintjes said. “Ten hailstones can show up as 1,000 raindrops.” The big advances since 2000, he explained, include dual-polarization radar, which emits wave signals on both the x and y axes, slicing into a cloud with amazing precision. “With dual-polarization radar, you can determine if it's hail or rain, and you can see the size and shape of the raindrop,” Bruintjes said. “It's really remarkable.”

Along with better data came increases in the power of computers to analyze the data and, more important, the ability to create virtual models to see what clouds would have done had they not been seeded. In October 2012 the NCAR switched on its Yellowstone supercomputer, a behemoth capable of 1.5 quadrillion calculations a second—180 times more powerful than the Bluesky supercomputer that the NCAR unveiled in 2002. Yellowstone lets Bruintjes and his colleagues assemble real-world data from the National Weather Service and a fleet of new NASA satellites—all of which are quite coarse—and create a numerical simulation of the cloud that is vastly finer. The computer can divide an area as big as 15 square miles into grid points as close together as 300 feet and carve up six hours of data into “time steps” of less than a second. This level of granularity delivers what Bart Geerts, a Belgian-born professor of atmospheric sciences at the University of Wyoming, called “the best representation of the atmosphere that we've ever had.” The computer is powerful enough, he said, to let scientists create a virtual sky: “There are a lot of idealized simulations—what-if simulations. You create a cloud, and you insert virtual silver iodide nuclei and see what happens.”

Only in the past year or so has Bruintjes been willing to be this declarative. “The evidence is strong,” he said, “that under certain conditions, we can increase rainfall by 10 to 15 percent.”

True Believers
Among the most enthusiastic believers in cloud seeding are West Texans, which is no surprise, given the seemingly perpetual drought under which they labor and the gigantic fires that sweep across their flat prairies, summer after summer. The West Texas Weather Modification Association has, since its founding in 1997, been charged with increasing rainfall over 6.4 million acres of southwestern Texas that lie within a zone that was getting only half its normal rainfall last summer. It is not doing research the way Bruintjes and Geerts are; it gets paid to put water on the ground.

The city of San Angelo and the water-conservation districts of seven counties pool $359,000 a year to support the association's effort, based partly on faith and partly on data, to squeeze a little more moisture out of the big, stubborn sky. It costs Texas and these farmers and ranchers, in other words, 4.4 cents an acre a year to gamble on the chance of making their fields a bit wetter. The dryland farmers want rain to fall directly on their crops. Irrigators and municipalities are eager to replenish the aquifers underneath the hardpan soil. What their money buys is the use of four single-engine airplanes, the part-time services of six retired military pilots at $75 an hour and a one-room office at the edge of San Angelo's sun-blasted airport. They also engage the full-time services of Jonathan Jennings, a 28-year-old meteorologist with a strong build and clipped-down hair who, at a time in which universities are churning out meteorologists, feels lucky to have a job. We met at the spare, unadorned office where he was keeping an eye on the computer monitor that feeds him 24 hours a day of data from Thunderstorm Identification Tracking Analysis and Nowcasting (TITAN), an NCAR program.

To me, the sky looked relentlessly cloudless, but Jennings was enthused about a couple of small, gray radar shadows he was watching over Crockett County; to him, they looked promising. Just outside the door sat one of Jennings's planes, an ordinary Piper Comanche, a low-wing four-seater that is a favorite of cloud seeders because its airframe is strong enough to withstand flying close to thunderstorms. The tips and trailing edges of its wings bristled with red-topped white tubes, each about a foot long and an inch around—the flares that apply the silver iodide to the clouds. Each was filled with Gilsonite, a type of flammable asphalt, mixed with 5.2 grams of silver iodate. When a pilot fires one, it burns hot and bright, transforming the iodate and leaving a trail of smoke containing iodide.

It was a hot, quiet day, occasional puffs of breeze stirring dust on the airstrip. I told Jennings I was a little disappointed. I had been expecting a scene from the Battle of Britain, with pilots hunkered in a ready room, drinking coffee and waiting to scramble while big, black thunderheads played the role of the German bombers. Jennings laughed. “You're not far off,” he said, “though we no longer keep the pilots hanging around.”

Every morning at about seven, Jennings e-mails his members and pilots a weather forecast assessing the likelihood of what he calls “operations.” Then he runs errands and goes to the gym, using his smartphone the whole time to monitor weather maps. Usually by around two in the afternoon, he knows if he is going to run seeding operations, and he will call the pilots to give them a heads-up. “When it's go time, we need to get them from phone call into the air in 30 minutes.”

Once the pilots have scrambled, things move swiftly, with Jennings watching his computer and acting as air-traffic control. “What I have to do is get them to the favorable part of the storm,” he said, which is the “inflow,” the tube of warm, moist air that rises into the storm and acts as its fuel. “Most of my pilots are experienced enough to know where the inflow is.” Sometimes you can even see it: ghostly tendrils of moisture rushing skyward. Pilots target the inflow because they cannot fly into the cloud. The wind shears inside could tear the plane to pieces, and Federal Aviation Administration rules forbid flying into thunderheads. And they no longer fly over the clouds because they discovered three drawbacks to doing so: it takes a lot of fuel to climb that high; the turbulence up there is brutal; and the chemical does not get deposited in the most efficient delivery zone—the inflow.

Instead Jennings's pilot circles the sweet spot, firing as many flares as he thinks necessary, letting the inflow carry the silver-laden smoke into the cloud. Sometimes one shot will do it; sometimes it takes as many as 50. Giving a cloud a silver lining takes 10 to 15 minutes. “The supercold water is about 2,000 feet inside the cloud,” Jennings said. The inflow carries the silver iodide up to precisely where it needs to go, causing the first ice crystals to form. “Once you trigger that reaction, the cloud naturally starts to create ice crystals. They start hitting each other and fracturing.” Each time an ice crystal fractures, it can pick up more moisture to carry earthward.

Jennings is experimenting with a new kind of flare that uses calcium chloride—salt—instead of silver iodide. Salt does not raise environmental concerns, it is cheaper than silver iodide (whose price is pegged to the price of mined silver and is now astronomical), and it works on warmer clouds and at lower relative humidity. In addition, some clouds seem to respond better to calcium chloride, Jennings said. On a few occasions, his pilots have deployed both. When that happens, so much rain pours out that “it's like dragging a knife along the underside of a cloud,” he noted. For him, there is nothing mystical about it. He does not have to count on long-term measurements and comparisons of what might have happened without his pilots: he sees the clouds respond instantly to their work.

“Look at this,” Jennings said and started playing the radar feed from April 28 on his computer. As we watched, tiny dots of yellow and pink—rain—twinkled in a few gray blotches. “When I saw that, I sent up the aircraft.” We followed the planes' flight paths inching across the screen. Within minutes of them reaching their targets, the yellow and pink dots swelled monstrously, smearing into a long line of boiling color. Cloud seeding not only creates droplets, Jennings explained, it can also lift clouds into a tall vertical structure that makes them “stronger,” as in better at producing rain. “We created a mesoscale squall line, an area of very strong convergence,” he said. “That gives more lift, which in turn gives more rain.” Of course, I responded, that might have happened anyway, without the seeding. He was ready for me: “The city of Sonora was predicted to have no rain that night and instead had an inch and a half.”

Vail, the Colorado ski resort, has been having the clouds above it seeded since 1975. Western Weather Consultants, a private contractor, operates 22 silver iodide generators on mountaintops in a 30-mile ring around the resort. When conditions are right, the generators, which are much less expensive to buy and operate than aircraft, are ignited to burn acetone permeated with silver iodide. The smoke rises into the clouds, and, the company says, as much as 35 percent more snow falls on the slopes than outside the target area. “Vail figures that cloud seeding for snow costs about 5 percent of what making snow costs,” Larry Hjermstad of Western Weather Consultants told me when I phoned him about it. Besides the generators at Vail, his company has 50 more operating up and down the continental divide for ski areas, municipalities and the states of the Colorado River basin. The regional drought that started three years ago has increased interest in cloud seeding, Hjermstad added, and with climate change and a rising population in the West, “we think of this as a long-term solution to a recurring problem.”

A Dubious History
The reputation of cloud seeding has not been helped over the years by the endless string of hucksters who tried to squeeze rain out of clouds and money out of suckers. As James Rodger Fleming of Colby College recounts in his dense and hilarious 2010 book, Fixing the Sky: The Checkered History of Weather and Climate Control, literature is full of weather changers, going back to the Bible and advancing forward through Jules Verne and, yes, Kurt Vonnegut. Serious “scientific” efforts to make rain go back to the mid-19th century, with people trying everything from cannon fire to forest fires to tickling clouds into raining. As recently as 1894, Nebraskans tried ending a hideous drought by touching off eight kegs of gunpowder at the Hastings fairgrounds. Typical of the science's ambiguity, a light sprinkle fell—not enough to do any good but just enough to encourage people to continue trying.

The federal government's scientific establishment used to be a devout believer; the National Science Foundation and the National Oceanic and Atmospheric Administration lavishly funded weather-modification experiments for 40 years, even in the face of setbacks. In 1962, for example, the government launched Project Stormfury: seeding hurricanes to reduce their intensity. A year later Category 4 Hurricane Flora killed thousands of people in Cuba, and Fidel Castro, still smarting from the previous year's missile crisis, accused the U.S. of manipulating the storm. But the government stuck with Stormfury for two more decades before conceding that seeding had no effect on hurricanes.

Just enough evidence trickled in, during the 1960s, to sustain belief that perhaps cloud seeding increases rain. The U.S. even employed it as a weapon in the Vietnam War. From 1967 to 1972, the air force seeded clouds over Laos in the hopes of slowing down North Vietnam's transport of men and materiel along the Ho Chi Minh trail, and it claimed to have increased rainfall by 30 percent. Although it was never clear why dropping rain on the enemy was more offensive than dropping napalm or high-explosive bombs, the revelation of “Operation Motorpool” in 1973 shocked the nation and the world. Cloud seeding began taking on malevolent connotations, and by 1977 the U.S. was compelled to sign an international treaty banning the manipulation of weather for military purposes.

Cloud seeding has proved contentious in other ways. On June 9, 1972, during a prolonged cloud-seeding experiment in South Dakota, a flash flood killed 256 people in Rapid City, and the ensuing lawsuit put the cloud seeders in the awkward position of arguing, essentially, the ineffectuality of their own undertaking. The case was dismissed on a legal technicality before the court could determine causality. Before and since Rapid City, farmers have been known to complain that seeding interfered with water that would have dropped on their farms had the clouds been left alone, and other, lesser floods have since been blamed on seeding. Cloud seeding's effectiveness has never been adjudicated, but the recurring incidents tarnish its reputation.

Then, of course, cloud seeders have had to contend with those who believe that they are interfering with God's plan; with those who think it is a capitalist plot to privatize the weather; and with those who are convinced that cloud seeding, crop dusting and even the contrails of high-flying jetliners are part of a “diabolical chemtrail genocide aerosol spraying operation” run by the government. One Web site, AboveTopSecret.com, describes how “Cloud Seeding Will Kill Us All.”

Some of the paranoia derives from the fact that silver iodide, a chemical used to develop photographs, is indeed toxic, especially to fish, and it is not just conspiracy theorists who worry about slinging the chemical around the atmosphere. Mainstream environmental groups have questioned the safety of cloud seeding since the 1970s, especially in light of seeding's dubious effectiveness. Francis Mangels, a former wildlife biologist for the U.S. Forest Service in California's Shasta-Trinity National Forest, has been fighting cloud seeding for years. “Silver iodide is an aquatic insect poison,” Mangels told a reporter in 2010. “Cloud seeding has never been adequately shown to work; it fails 95 percent of the time, and it's poison. Doesn't that say it all?”

Not quite. The truth is that, though toxic, silver iodide is applied in such tiny quantities as to be all but impossible to measure in the environment. The kind of clouds that are ripe for seeding generally contain between 10,000 and 30,000 kilotons of water, so the 40-odd grams of silver iodide used in a typical seeding is infinitesimal. Altogether cloud seeding worldwide annually constitutes about a tenth of a percent of the total silver that human activity in the U.S. adds to the biosphere. The cloud-seeding industry continues to argue that the silver iodide it uses is not detect-able above background levels in either soil or groundwater and that it poses no threat to either humans or fish, although that does not mean it can expect the issue to go away anytime soon.

Scientists believe that it was the combination of controversy and uncertain results that led the federal government to pull out of weather-modification research in the 1980s. Bill Woodley, a retired meteorologist now on the board of the Journal of Weather Modification, recalled running a promising cloud-seeding experiment in Florida in the 1970s that suddenly lost its funding. Although he and his colleagues seemed to have increased rainfall in their 13,000-square-kilometer area by about 15 percent, they had predicted more. “Some people in the media said, ‘Well, then, it was a failure.’ We tried to say, ‘No, we learned a lot,’ and put in for funding for a confirmatory phase,” but noaa pulled the plug, Woodley asserted. “People were saying, ‘If it isn't obvious and provable, we don't need the grief.’”

To scientists, promising but ambiguous data are an argument for more intense investigation, not less. “A reasonable scientist would say, ‘It's clear [cloud seeding] works under some circumstances, but how often do those circumstances occur in an area that makes economic sense, and how do you quantify it on the ground?” recounted Dan Breed, another NCAR meteorologist. To government officials making funding decisions in a highly charged political environment, though, conflicting data have been an excuse to withdraw from an increasingly controversial enterprise. “At a certain point, the [federal] government said, ‘To hell with it; it's not worth it,’ and got out of funding research altogether” in the early 1980s, said Joseph Golden, who once chaired noaa's now defunct Atmospheric Modification Program and who today works with the Utah-based Weather Modification Association, a consortium of 18 Western cloud-seeding projects. Golden is a jolly and ruddy man in his 60s. When we met for coffee in Boulder, Colo., he was ready to talk for hours about the fecklessness of the federal research establishment. “We need to have a neutral evaluator [of the data]. That would be the government. But there is no federal presence in supporting research because it is controversial.” So for the past 20 years, scientists studying weather modification in the U.S. have done so without federal funding.

Those keeping track of the ways that the Chinese are surpassing Americans can add public support for cloud seeding to the list. The Chinese government's weather army has a goal of squeezing 3 to 5 percent more rain out of the sky during this decade. They claim to have generated almost 500 billion tons of rain that otherwise would not have fallen. Thailand has been seeding clouds since the 1960s—using a method patented by King Bhumibol Adulyadej himself called “super sandwich,” which calls for simultaneously seeding both warm and cold clouds floating by at different levels. (It is hard to find reliable information about the program's effectiveness, though, because it is a crime to say anything negative about the king in Thailand.)

Malaysia aggressively seeded clouds this year to make rain, which seems to have also fallen on neighboring Singapore. This year Indonesia, which experimented two years ago with cloud seeding to reduce haze from forest fires, seeded clouds to divert flooding rains from Jakarta. Russia is a big believer in the technology and deployed it to wash radioactive particles out of the air after the Chernobyl nuclear meltdown in 1986. In all, 50 countries participate in cloud seeding, most of them with the assistance of Bruintjes and his colleagues at the NCAR.

Enhancing the Rain
Airplanes, to say nothing of rocket launchers and antiaircraft guns, are blunt instruments. They are expensive to operate and maintain, they pollute, and the whole cloud-seeding process can seem hopelessly 20th century. So it is no surprise that people are searching for cleaner, more advanced ways of generating rain. The current fad is for ionizing the atmosphere; in the laboratory, filling the air with charged particles causes moisture to clump and fall. A project in Abu Dhabi fields antennas that look like gigantic umbrella frames, and the project's scientists claim to be yielding results, as do others in Australia, whose antennas look like Brobdingnagian jungle gyms. Not a single scientist interviewed for this story had much faith in either the theory or the practice of making rain by ionizing the atmosphere, though. Bruintjes went so far as to call it “fraud.” Even one of the first scientists to experiment with the technique, Arquimedes Ruiz-Colombié, who in the 2000s ran an experiment in Laredo, Tex., that attempted to make rain with an ionization antenna the size of a circus tent, found no proof that it produced rain.

Ruiz-Colombié is a large, jovial 61-year-old who began his career in Cuba before being imprisoned and then expelled for political activity in the 1990s. Now an instructor at Texas Tech University, he works with Jennings at the San Angelo seeding project. While Jennings and I were talking, he came thundering into the office. He told me emphatically that despite what I might have heard, his ionization experiment was not a failure. It just had different results than anticipated. “We found no signal for increased rain—that's true,” he said. “But what we did find is that downwind of the tower, the concentration of aerosols [airborne particles] was less. They stick together and fall to the ground. So ionization cleans the environment.” As for the Abu Dhabi and Australia rainmaking experiments, Ruiz-Colombié was “very skeptical. But I have an open mind,” he said. “Show me the data.”

With Ruiz-Colombié nodding modestly beside him, Jennings explained that as much as anything, Ruiz-Colombié's meticulous data collection, mathematical work and modeling of cloud behavior have demonstrated the validity of cloud seeding. They handed me a 10-year analysis of the efforts of their parent organization, a statewide weather-modification association that covers 35 counties. The 3,100 seeded clouds in the study grew larger and lived longer than unseeded clouds outside the target area—and they dumped a total of 3.4 million more acre-feet of water, almost 12 percent more than the unseeded clouds.

“Understand something, please, because this is what you call the bottom line,” Ruiz-Colombié said, sitting forward and holding up a finger. “We cannot ‘make’ it rain. If there are no clouds or not the right clouds, we cannot make something out of nothing. What we do is enhance rain.”

“Right,” Jennings chimed in. “Think of pulling a sponge out of a bucket of water. You can hold it up and let it drip, or you can squeeze it. What we do is squeeze.”

*Erratum (6/27/14): This sentence incorrectly states that the volume of a small cloud can top 750,000 cubic kilometers. The correct volume is 750,000 cubic meters.