It sounds like science fiction: A superstorm barrels toward the coast, and instead of bracing for impact, scientists give the atmosphere a gentle shove days in advance — nudging the storm off course before it ever makes landfall.

That's the incendiary idea behind a new paper in the journal PLOS Water, in something the researchers have coined: "weather jiu-jitsu."

The concept borrows from the martial art, where you don't overpower an opponent, but use their own momentum against them. Similarly, the authors argue, it would be far easier to nudge the steering winds that guide the storm, rather than trying to drain a hurricane's enormous energy.

Weather in your inbox

Forecast Location

Sign Up

By signing up you agree to the Terms & Privacy Policy. Unsubscribe at any time.

But before anyone pictures a giant off-switch for bad weather, here's the kicker: None of this has happened. At least, not in the real world. The findings come entirely from computer simulations, and the idea is already drawing sharp pushback from scientists who warn that "controlling the weather" is a door we may not want to open.

Storm 'nudging' simulated with AI

A team led by Qin Huang at Arizona State University used a cutting-edge artificial-intelligence weather model to replay three real disasters and asked what might have happened if they had nudged the atmosphere first.

Here are the three examples the team used:

Get More of Our News on Google

Set The Weather Company as a 'Preferred Source' to get quicker access to the news you value.

• Superstorm Sandy in 2012: A nudge seven days before landfall shifted the track by about 200 miles.

• Texas freeze in 2021: Tweaking the upstream pattern eased the worst of the cold by roughly 18 degrees Fahrenheit.

• A California atmospheric river in 2022: A single nudge trimmed the moisture flooding into the West Coast by about 5%.

How would this work in real life?

The "nudge" itself would be delivered by cloud seeding. Cloud seeding has been around for decades and can be quite controversial. Planes or ground generators release particles like silver iodide to coax a little more rain or snow out of clouds that are already there.

China, the United Arab Emirates and several U.S. states use it.

What it does not do (despite what viral posts after every flood claim) is conjure storms out of nothing or "control the weather."

This study takes that same tool and points it at a far bigger target.

Instead of seeding clouds to make more rain, the idea is to use those small, seeding-induced nudges to steer entire weather systems: moving a hurricane, softening a cold snap, weakening a firehose of Pacific moisture. It's the same technique, just a vastly bigger ambition, and one that raises even bigger questions.

One of the biggest caveats when it comes to altering the weather is that you have to have existing clouds in order for cloud-seeding to be effective. Without the moisture and clouds present, it won't work. Therefore, examples using cold air especially would be much more difficult to pull off.

The appeal, the authors say, is leverage: In their idealized models, the energy required stayed below 2% of the storm system's own energy.

Not everyone buys the math. Tom Hamill, weather.com's Senior Director of numerical weather prediction, says the study's biggest selling point — that a nudge would take less than 2% of a storm's energy — is misleading once you scale it to a real hurricane.

By his own calculations, even a small fraction of the energy in a typical midlatitude storm is "comparable to the energy released by a megaton nuclear warhead."

And unlike a bomb, he notes, that energy would have to be spread across an area the size of the storm itself. Worse, cloud seeding only works where the atmosphere is already primed for towering thunderstorms, which may be nowhere near the spot a nudge is actually needed.

The ethical dilemma

If you can steer a storm away from one place, you are, by definition, steering it toward somewhere else.

If you move a Sandy-like hurricane 200 miles off New Jersey, it comes ashore on someone else's coastline. Weaken an atmospheric river bound for California, and you may rob a drought-stricken region of water it desperately needed.

The authors even acknowledge that such interventions "can create winners and losers across national boundaries."

So, who decides which town gets spared and which gets sacrificed? Who's liable when the nudge goes wrong? Or when a disaster strikes a region that suspects it was pushed their way?

It's not just hypothetical, Hamill says.

Take the very Sandy scenario above: Nudge the storm away from New Jersey and you may simply hand it to Boston.

"The people in Boston wouldn't be happy," he says.

Hamill is also wary of the model itself. The calculations that pinpoint those crucial "leverage points," he says, hinge on a single model run and its exact starting conditions. Feed in slightly different inputs, as forecasters routinely do when they run dozens of versions of a model, and the answer could change entirely. In a system as chaotic as the atmosphere, that's a shaky foundation to bet a city on.

A long way from the sky

Remember, this isn't something that will be launched with the next hurricane. This is simply a "proof of concept," at this point. There were just three historical cases re-run inside a single AI model. And while the model can show that a simulated storm can be diverted, it doesn't mean we could pull it off in the real atmosphere, or that we should.

Even the researchers acknowledge the idea would require major advances in monitoring, attribution and "international governance" before it could ever leave the lab.

For now, "weather jiu-jitsu" is a bold experiment that will most likely spark debate. And it's also a reminder that as tools used to model the atmosphere get more powerful, the questions about whether to meddle with it are only going to get harder.

For all the buzz, the core idea isn't new. A strikingly similar proposal surfaced in the scientific literature more than 20 years ago, and the same objections were raised then.

Hamill's verdict today: "I don't see any practical path forward for this technology in the near future, given the amount and distribution of energy that would be [needed] and the underlying scientific challenges."