There are a couple different ways to come at the problem of climate change—you can focus on eliminating the cause, or on mitigating the symptoms. The latter approach includes obvious things like preventing flooding from rising sea levels. But it also ranges into “geoengineering” schemes as radical as injecting sunlight-reflecting aerosol droplets into the stratosphere. Such schemes are band-aids rather than cures, but band-aids have their uses.

One worrying change driven by the climate is the loss of Arctic sea ice. The late-summer Arctic Ocean is on track to become ice-free around the 2030s. The rapid warming of the Arctic has serious implications for local ecosystems, but it also influences climate elsewhere in ways we’re still working to fully understand. One frequently mentioned effect is the increased absorption of sunlight in the Arctic as reflective snow and ice disappears—a positive feedback that amplifies warming.

What if we could slap a sea ice band-aid on the Arctic? In a recent paper, a group of Arizona State researchers led by astrophysicist Steven Desch sketch out one hypothetical band-aid—a geoengineering scheme to freeze more ice during the Arctic winter.

The idea is simple enough: wind turbines on the sea ice could pump water from below onto the surface, where it quickly freezes, thickening the ice in winter. In the right places, that could mean the difference between sea ice disappearing or surviving through the end of summer. But like any back-of-a-napkin solution to the world’s problems, reality is substantially more challenging than it might initially appear.

A good chunk of the paper is dedicated to the physics of freezing seawater. While the Arctic Ocean is actually slightly colder than a bucket of ice water, the saltiness of seawater lowers the freezing point to about -1.8°C. Because the air above the Arctic Ocean can be much, much colder than that, ice seasonally forms at the ocean’s surface.

Existing sea ice forms a barrier between the seawater and the frigid air, and any new ice forms on the bottom of that ice sheet. So thickening is inhibited by the need to transfer the heat released during freezing from the bottom of the ice to the atmosphere. Though the wind turbine scheme brings water up to freeze on top of the ice, the thickening at the bottom is still a significant factor. The researchers calculate that pumping enough water to add a meter of ice would actually only result in a net thickening of 0.7 meters because the “natural” addition of ice to the underside would be reduced.

That aside, could a scheme like this actually halt the loss of sea ice? The annual average thickness of sea ice in the Arctic, which is now about 1.4 meters, is decreasing by almost 0.6 meters per decade. The researchers focus on a scenario where their turbines cover 10 percent of the Arctic ice, thickening it by a meter over each winter. Because the additions can carry over somewhat from year to year if you choose your locations well, this would actually be more than enough to counter the downward trend.

Geoengineering doesn’t come cheap, and this is no exception. As the researchers put it, “[I]t is reasonable to ask whether such an endeavor is financially feasible or even logistically possible.” Covering “just” 10 percent of the Arctic would require a staggering 10 million wind turbine pumps like those currently used on farmland. Mounting 12-meter-tall turbines on steel buoys brings you to about 10 tons of steel each.

Throw in the cost of shipping these things off to their Arctic homes, and setting up the turbines over 10 years would cost about $50 billion per year. That’s certainly no pittance, but it’s also far less than the US paid for the Iraq War, for example. Of course there are also maintenance costs—it’s unclear how easy it would be to ensure the turbines don’t simply become encased in the ice they’re trying to make. The authors don’t calculate this expense

There are, obviously, many blanks to be filled in before a proposal like this could be properly evaluated. Maintenance costs—and how best to prevent the turbines from becoming encased in ice and rime—are just one unknown. Would turbines need to be periodically relocated as they drifted with the sea ice? And the ice itself might be different. Since seawater is frozen on top of the ice, the brine formed from the salt excluded by the freezing process would burrow downward, changing the physical properties of the ice in a way that might have consequences.

Stabilizing sea ice in some areas could have clear benefits for the species that rely on it, including polar bears. The climate benefits could be significant, too, holding off the amplifying feedback of absorbing more sunlight. That avoided warming could partly prevent the release of greenhouse gases from thawing permafrost or possible changes in weather patterns driven by sea ice loss.

The cost-benefit calculation is exceedingly complex because the Earth’s climate system is complex. Along with the incredible scale, that is what makes geoengineering a daunting proposition. Then again, the future without it is looking daunting as well.

Open Access at Earth’s Future, 2016. DOI: 10.1002/2016EF000410 (About DOIs).