David Mitchell pulls into the parking lot of the Desert Research Institute, an environmental science outpost of the University of Nevada, perched in the dry red hills above Reno. The campus stares over the tops of the downtown casinos into the snow-buried Pine Nut Mountains. On this morning, wispy cirrus clouds draw long lines above the range.

Mitchell, a lanky, soft-spoken atmospheric physicist, believes these frigid clouds in the upper troposphere may offer one of our best fallback plans for combating climate change. The tiny ice crystals in cirrus clouds cast thermal radiation back against the surface of the earth, trapping heat like a blanket—or, more to the point, like carbon dioxide. But Mitchell, an associate research professor at the institute, thinks there might be a way to counteract the effects of these clouds.

It would work like this: Fleets of large drones would crisscross the upper latitudes of the globe during winter months, sprinkling the skies with tons of extremely fine dust-like materials every year. If Mitchell is right, this would produce larger ice crystals than normal, creating thinner cirrus clouds that dissipate faster. “That would allow more radiation into space, cooling the earth,” Mitchell says. Done on a large enough scale, this “cloud seeding” could ease global temperatures by as much as 1.4 °C, more than the planet has warmed since the Industrial Revolution, according to a separate Yale study. 

Big questions remain about whether it would really work, what damaging side effects might arise, and whether the world should risk deploying a tool that could alter the entire climate. Indeed, the suggestion that we should entrust the global thermostat to an armada of flying robots will strike many as preposterous. But the real question is: preposterous compared to what?

Without some kind of drastic action, climate change could kill an estimated half-million people annually by the middle of this century, through famine, flooding, heat stress, and human conflict. Preventing temperatures from rising 2 °C above preindustrial levels, long considered the danger zone that should be avoided at all cost, now looks nearly impossible. It would mean cutting greenhouse-­gas emissions by as much as 70 percent by 2050, and it may well require developing technologies that could suck billions of tons of carbon dioxide out of the atmosphere, according to the U.N.’s Intergovernmental Panel on Climate Change. But a growing body of research suggests that we probably will not have the time or technology to pull this off. Notably, even if every nation sticks to the commitments it’s made under the politically ambitious Paris climate accords, global temperatures could still soar more than 5 °C by 2100.

“Everyone is looking at two degrees, but to me it’s a pipe dream,” says Daniel Schrag, director of the Harvard University Center for the Environment, who was one of President Obama’s top advisors on climate change. “I fear we’ll be lucky to escape four, and I want to make sure nobody ever sees six.”

The difference between two and four degrees is another quarter-billion people without reliable access to water, more than a hundred million more exposed to flooding, and massive declines in worldwide crop yields, according to a study by the Committee on Climate Change, a London-­based scientific group established to advise the U.K. government (see below).

The idea that we could counteract these dangers by reëengineering the climate itself, techniques collectively known as geoengineering, began to emerge from the scientific fringes about a decade ago (see “The Geoengineering Gambit”). Now momentum behind the idea is building: increasingly grim climate projections have convinced a growing number of scientists it’s time to start conducting experiments to find out what might work. In addition, an impressive list of institutions including Harvard University, the Carnegie Council, and the University of California, Los Angeles, have recently established research initiatives.

Few serious scientists would argue that we should begin deploying geoengineering anytime soon. But with time running out, it’s imperative to explore any option that could pull the world back from the brink of catastrophe, says Jane Long, a former associate director at Lawrence Livermore National Laboratory. “I don’t really know what the answer is,” she says. “But I do believe we need to keep saying what the truth is, and the truth is, we might need it.”

Dreams of dust

Mitchell works in a small, square office on the top floor of the Desert Research Institute. Stacks of scientific papers crowd his desk; journals and binders pack his bookshelf. Close-up images of delicate ice crystals hang from thumbtacks on the bulletin board above his computer monitor.

In the spring of 2005, during a sabbatical at the National Center for Atmospheric Research in Boulder, Colorado, Mitchell began exploring how the size of ice crystals affects cirrus clouds and the climate system. He and his colleagues found that bigger crystals, the type that tend to form in the presence of dust particles, produced fewer and thinner cirrus clouds.

That point stuck in Mitchell’s brain. One morning shortly after returning to Nevada, he had a dream in which that insight morphed into a climate engineering scheme. He awoke wondering if deliberately adding dust in the areas where these clouds form would spawn these larger ice crystals, reducing cirrus coverage and releasing more heat into space.

Though he had serious reservations about geoengineering, he decided to explore the idea. In 2009, he and a colleague published a paper suggesting that seeding cirrus clouds with tiny particles of bismuth tri-iodide, an inorganic compound that may break down into the necessary sub-micrometer size, might substantially offset climate change. More recently, Mitchell estimated that it would take around 160 tons of the material annually to seed clouds in the areas he has in mind, at a cost of about $6 million.

Not everyone agrees the proposal would work. A 2013 paper in Science, led by MIT atmospheric scientist Dan Cziczo, concluded that the formation of ice crystals around dust, known as heterogeneous ice nucleation, is already the dominant mechanism creating cirrus clouds. That might mean adding more dust would, on balance, create thicker clouds that trap more heat. The larger problem with the idea, Cziczo argues, is that clouds are the least understood part of the climate system. We do not have nearly enough knowledge about cloud microphysics, or accurate enough measurements, to precisely manipulate climate in this way, he says.

But Mitchell’s most recent research, relying on observations of ice crystal concentrations from NASA’s Calipso satellite, has further convinced him that cloud seeding could work, as long as it’s done in regions where cirrus clouds form primarily without dust particles. On the monitor in his office, Mitchell pulls up a page of maps from a paper he presented at the National Center for Atmospheric Research in late February. Navy- and light-blue dots, representing Cziczo’s heterogeneous clouds, dominate the mid-latitudes, covering much of South America and Africa. But the higher latitudes are covered in red, yellow, orange, and green dots that indicate the sorts of clouds Mitchell has in mind.

The satellite images suggest that in very cold and humid conditions, toward the poles and particularly during winter, tiny ice crystals can form on their own, spontaneously, without dust. That suggests that cloud seeding could work, if it’s targeted to those areas during those months. Mitchell even thinks he’s come up with a way to get nature to carry out a field experiment to test his theory. During spring and winter, strong winds regularly stir up major dust storms in the deserts of Mongolia and the western edge of China. The fine particles blow across the Pacific and run into an atmospheric wave that rolls over the Rocky Mountains.

If Mitchell is correct, the dust should promote thinner cirrus clouds in an area where the thicker type otherwise tends to dominate. There was no way to properly observe this phenomenon—until late last year, when the National Oceanic and Atmospheric Administration launched a satellite equipped with some of the most powerful imaging technology ever launched into space, as well as sensors that can measure the temperatures of clouds. The satellite should be able to capture exactly what happens as the dust rides over the Rockies, detecting the subtle shifts under way in cloud microphysics.

Mitchell submitted a research proposal to NOAA last year, asking the agency to use the satellite to make such observations. He knows it’s a long shot, particularly in light of the Trump administration’s efforts to slash funding for climate science. But if NOAA agrees, the test could lend weight to his theory—or, of course, contradict it.

Another outdoor geoengineering experiment should occur even sooner.

By this time next year, Harvard professors David Keith and Frank Keutsch hope to launch a high-altitude balloon from a site in Tucson, Arizona. This will mark the beginning of a research project to explore the feasibility and risks of an approach known as solar radiation management. The basic idea is that spraying materials into the stratosphere could help reflect more heat back into space, mimicking a natural cooling phenomenon that occurs after volcanoes blast tens of millions of tons of sulfur dioxide into the sky (see “A Cheap and Easy Plan to Stop Global Warming”).

Scientists generally believe the technique would ease temperatures, but a lingering question is: what else will it do? Notably, volcanic eruptions have also significantly altered rainfall patterns in certain areas, and sulfur dioxide is known to deplete the protective ozone layer.

Keith has done extensive climate modeling to explore whether other materials, including alumina, diamond dust, and calcium carbonate, might have a neutral or even positive impact on ozone. During a conversation in his office at Harvard, he stressed that the experiments wouldn’t constitute a test of geoengineering itself. But they would allow his group to subject its models to real-world data, revealing more about the relevant stratospheric physics and chemistry. “Theory alone doesn’t tell you what will happen in the atmosphere,” Keith says. “You can fool yourself if you don’t go out and make direct measurements.”

Keith has already begun design work with the balloon company World View Enterprises, as well as discussions about the appropriate transparency and oversight for such outdoor experiments. The early flights would test the basic workings of the balloon, which would be tethered to a gondola equipped with propellers, sprayers, and sensors. But eventually the experiment would involve releasing a fine plume of materials, probably calcium carbonate, into the stratosphere. The balloon would then track that trail in reverse, allowing the sensors to measure how well the particles scatter sunlight, whether they coalesce or disperse, and how they interact with precursors to ozone.

Unknown unknowns

Full-scale geoengineering would inevitably involve some level of risk. We are likely to face a terrible choice between accepting the clear dangers of climate change and risking the unknowns of geoengineering. Alan Robock, a professor of environmental sciences at Rutgers, has published a list of 27 risks and concerns raised by the technology, including its potential to deplete the ozone layer and to decrease rainfall in Africa and Asia.

Ultimately, Robock worries that geoengineering may simply be too risky to ever try. “We don’t know what we don’t know,” he says. “Should we trust the only planet known to have intelligent life to this complicated technical system?” MIT’s Cziczo is blunter. “We know the problem is greenhouse gas, so the solution is you take the greenhouse gas out,” he says. “You don’t try to do something that we completely don’t understand.”

The reservations surrounding geoengineering research were on full display in late March as dozens of notable climate and social scientists gathered at the Carnegie Endowment for International Peace in Washington, D.C., for the Forum on U.S. Solar Geoengineering Research. Speakers highlighted a long list of unanswered, and perhaps unanswerable, questions about international governance: Who gets to decide when to pull the trigger? How do we determine “correct” average temperatures when the same ones will affect different nations in markedly different ways? Can one nation be held responsible for the negative effects of its geoengineering scheme on another country’s weather? Could these tools be used to deliberately attack a neighboring nation? And could conflicts over these questions tip into war?

“I have yet to hear any description of a future solar-geoengineered world that sounds to me anything other than dystopian or highly unrealistic,” said Rose Cairns, a research fellow at the University of Sussex, who joined the morning discussion from England by Skype.

But Harvard’s Schrag argued the opposite: that the scariest version of the future may be one where geoengineering is never developed or deployed. “I don’t think people understand just what we’re up against with climate,” he said. “The most likely scenarios for climate over longer time scales are devastating to future generations, absolutely devastating.”

As he flashed slides highlighting the dramatic loss of sea ice in the Arctic and Antarctic in recent months, Schrag stressed that climate change is already causing visible impacts faster than anyone expected. He added that it’s difficult to foresee any scenario where we can cut greenhouse-­gas levels fast enough to avoid far worse dangers: the amount we’ve already released is likely to lock in another degree of warming even if we halt emissions tomorrow, he said.

To his mind, these hard realities mean we need to try to answer the difficult questions geoengineering poses. “It is still, in every case that I’ve seen, better than the alternative of just letting the climate warm,” he said. “Given the trajectory of the world, and the difficulty of reducing emissions, this is something we really need to understand.”

The power of fear

Mitchell was opposed to geoengineering for most of his career. The idea that humankind should tinker with the finely tuned climate system struck him as impossibly arrogant. But like other researchers who spent decades staring at increasingly frightening projections while the world ignored the loudest warnings scientists knew how to sound, he reluctantly changed his view.

It could take decades to learn which geoengineering methods might work, whether environmental side effects can be minimized, and whether it’s ultimately too dangerous to try. The longer we wait to begin serious research, the greater the risk we’ll deploy an unsafe tool in the face of sudden climate shocks, or not have one in hand when we need it. And no one really knows when that might be.

Says Mitchell, “The need for climate engineering could be coming faster than we realize.”