Arizona State University scientists are exploring whether artificial intelligence and precise atmospheric interventions could help redirect hurricanes and other extreme weather events away from populated areas. The approach, called "weather jiujitsu," borrows from martial arts philosophy: using small, strategically placed nudges to redirect massive forces rather than trying to overpower them directly.

What Is Weather Jiujitsu and Why Does It Matter Now?

The concept centers on identifying moments when atmospheric systems like jet streams become unstable, then applying minimal perturbations to guide them in safer directions. Upmanu Lall, director of the ASU Water Institute and a professor in the School of Complex Adaptive Systems, explained the reasoning: "We don't want to control the atmosphere all the time, because that would make it a very boring planet. But we want to control it so that when there is a weather event that is likely to be very impactful, we can steer it away, towards a harmless area. That's weather jiujitsu".

"We don't want to control the atmosphere all the time, because that would make it a very boring planet. But we want to control it so that when there is a weather event that is likely to be very impactful, we can steer it away, towards a harmless area," said Upmanu Lall.

Upmanu Lall, Director of the ASU Water Institute and Professor at Arizona State University

The urgency is real. Much of the United States' weather-protective infrastructure, including dams and levees, has exceeded its design lifespan. Combined with climate change intensifying extreme weather events, communities face growing exposure to hurricanes, atmospheric rivers, droughts, deep freezes, and heat waves.

How Could AI Help Predict and Control Extreme Weather?

The research team used Aurora, an artificial intelligence weather forecasting model, to test their concept. In a simulation of Hurricane Sandy, Aurora correctly predicted the storm would hit New York City and New Jersey five days before landfall. Researchers then identified theoretically suitable locations where small perturbations could alter the storm's track. When they simulated cloud seeding at those locations, the resulting wind changes diverted Sandy out to sea, keeping the region safe.

AI-based weather models are proving significantly more accurate than traditional forecasting methods, especially for longer-duration predictions. The computational speed is another major advantage. "It's very easy to run those forecasts with AI. The computation is very fast, especially with the ASU Sol supercomputer. The best part about AI is that it accelerates everything and improves efficiency while keeping comparable accuracy for extreme weather prediction," explained Qin Huang, an ASU graduate student involved in the research.

"It's very easy to run those forecasts with AI. The computation is very fast, especially with the ASU Sol supercomputer. The best part about AI is that it accelerates everything and improves efficiency while keeping comparable accuracy for extreme weather prediction," said Qin Huang.

Qin Huang, Graduate Student at Arizona State University

What Intervention Methods Could Trigger Weather Changes?

Cloud seeding, which has been used for 80 years, remains one proven mechanism. When rain forms through cloud seeding, it releases terawatts of energy into the atmosphere. But researchers are exploring additional approaches with similar energy-release potential.

One intriguing concept involves specialized laser technology. Researchers would point multiple lasers into a cloud to focus them at a single intersection point. Lightning in the atmosphere would then discharge into that spot, channeling through the laser beam into the ground. "Again, it's jiujitsu, because we are taking a simple light beam, and we could leverage it to create up to a terawatt hour of energy discharge," Lall noted.

• Cloud Seeding: A proven 80-year-old technique that disperses salt or silver iodide into clouds to help water droplets or ice crystals form, releasing massive amounts of atmospheric energy.
• Laser-Triggered Lightning Discharge: Focused laser beams intersect in clouds to direct atmospheric lightning into a controlled ground discharge, potentially releasing up to a terawatt hour of energy.
• Precision Perturbation: Small nudges applied at moments of atmospheric instability, leveraging the natural energy of weather systems to redirect their paths safely.

Could These Interventions Accidentally Make Things Worse Elsewhere?

Yes, and researchers acknowledge this risk directly. Lall stated: "Definitely, it can. That's why the precision control of jiujitsu is important". This concern is why the team is breaking their work into two phases. First, they want to demonstrate that precise control is actually possible. Only after proving that capability would they move to the second phase: determining how to deploy these interventions responsibly.

Lall

The control mechanism would need to be continuous, similar to how autonomous vehicles operate. Every intervention would be observed, evaluated, and adjusted in real time. If results deviate from expectations, corrections would be applied immediately.

What Are the Next Steps for This Research?

Researchers plan to test their approach in safe environments first, such as over the ocean where no people live. If a storm is predicted to remain over open water, scientists would attempt to disrupt it there and observe whether they can control its behavior. Only after demonstrating success in these controlled settings would the team consider actual field deployment.

The broader context for this work extends beyond hurricanes. In Arizona, researchers are considering whether weather jiujitsu could help prevent droughts. The difference between a drought year and a normal year in the upper Colorado Basin may come down to just 4 to 5 additional storms during the season. If targeted interventions could reliably produce those extra storms, severe droughts might be avoided or at least made more predictable.

How Is AI Transforming Water Security More Broadly?

While weather control remains experimental, AI is already reshaping how communities manage water resources. Water security is emerging as one of the defining challenges of this century, driven by drought, industrial demands, climate variability, aging infrastructure, and emerging contaminants.

AI applications in water systems include predictive water quality management, which uses machine learning to detect early trends before operational parameters deviate from regulatory thresholds. Predictive maintenance and digital twins also help forecast equipment failures and simulate scenarios to strengthen resilience.

In desalination, AI analyzes feedwater conditions, pressure, recovery rates, and energy prices in real time to reduce electricity consumption while maintaining steady freshwater output. AI also improves membrane management by predicting fouling from biological growth and scaling, enabling proactive cleaning before performance drops.

For complex contaminants like PFAS (per- and polyfluoroalkyl substances), AI helps utilities optimize treatment selection by analyzing water chemistry, flow rates, and treatment goals to predict performance across different technologies. It also improves adsorption system management by predicting media breakthroughs, reducing unnecessary replacement costs while protecting water quality.

Despite these technological advances, human expertise remains irreplaceable. Water security will not come from a single technology alone. It requires better data, stronger decisions, and operational excellence, with skilled professionals using AI as a powerful partner to build safer, more reliable, and more sustainable water systems.

" }